Method for high-throughput cheese making

FIELD OF THE INVENTION

The present invention relates to a method for producing cheese. The method is particularly useful in high-throughput cheese making applications. The present invention also relates to the screening of the cheese produced by said method, the screening particularly relating to the characterisation of one or more features which are characteristic of, or which are determining for one or more organoleptic properties, health properties and/or food safety aspects of the cheese.

BACKGROUND OF THE INVENTION

Cheese is an important product, both in terms of nutritional and economical value. For the consumer of cheese, factors of importance include (1) the quality of the cheese, particularly in terms of its flavour, taste, healthiness and nutritional value, (2) the availability of different types of cheese, each having its particular organoleptic properties, and (3) the assurance of food safety, particularly in that pathogenic microorganisms are absent from the cheese.

These factors are dependent on a complex interplay of cheese making ingredients and cheese making process steps.

Important cheese making ingredients are milk, one or more coagulants (such as chymosin, calf rennet, or microbial coagulant), and one or more bacterial cultures.

Many important commercial (traditional) cheese making processes comprise the following sequential process steps:

^■ renneting a composition comprising milk, at least one coagulant and at least one bacterial culture;

^■ cutting the obtained curds; ^■ optionally washing the curds;

^■ removing excess liquid from the curds, for example by pressing; and

^■ ripening the curds.

In addition, a brining step is commonly included. For example, in the commercial production of Gouda-type cheese, brining occurs by a process wherein - prior to the ripening step - blocks of pressed curds are placed in a concentrated aqueous solution of sodium chloride.

The criteria of quality, diversity and food safety will now be briefly addressed, with a focus on the influence of the bacterial culture.

The bacterial culture may be in the form of a starter culture. The starter culture usually comprises a mixture of bacteria, such as various species of the Lactobacillus genus or of the Lactococcus genus, and/or mixtures thereof. One function of the starter culture is to provide for acidification of the milk, i.e. to convert lactose into lactic acid. This function is fulfilled within the first 48 hours of the cheese-making process. Secondly, the starter culture determines the general organoleptic properties of a certain cheese type. For example, the production of traditional Gouda-type cheese always requires the use of specific mesophilic starter cultures. As will be discussed in more detail below, the organoleptic properties of the cheese are slowly developed over time, during the ripening process.

The bacterial culture may also be in the form of an adjunct culture, which usually involves a single bacterial strain. Typically, the cheese maker employs a starter culture in combination with an adjunct culture. The main role of the adjunct culture is usually to impart a certain desired flavour and/or taste to the cheese. For example, the use of a thermophilic adjunct culture in a Gouda-type cheese making process typically provides for a Gouda-type cheese having additional sweet notes, which are reminiscent of the flavour of certain hard-pressed Italian cheeses.

It is known that certain proteolytic enzymes are produced by and/or liberated from the bacterial cultures during ripening. These bacterial enzymes are important for the organoleptic properties of the cheese by breaking down the casein to form small peptides and volatile aroma components, thereby affecting the consistency, the taste and the flavour of the cheese as developed during ripening.

Furthermore, it is noted that enzymatic activity from enzymes as naturally present in (raw) milk, and/or from added functional ingredients (such as lipolytic enzymes) also contributes to the development of one or more organoleptic properties of the cheese during ripening.

It follows that to ensure a consistent quality of the cheese, each batch of starter cultures should contain a consistent and reproducible population of bacterial strains, each of which should have a consistent and reproducible activity in the cheese. The criteria of consistency and reproducibility of the starter culture are commonly checked by the producer of said culture after the production of each batch.

The diversity of the various cheese types can be determined by the use of (mixtures of) adjunct cultures, proteolytic enzyme preparations, and/or functional ingredients in the preparation of the cheese milk composition. The extent to which the flavour profile of the cheese type is influenced, also strongly depends on the conditions employed in the cheese making process.

As healthiness of the cheese is a growingly important aspect, the composition of various cheese types is accordingly adapted, preferably in the direction of lower salt and/or fat content. These compositional changes usually have a significant impact on the organoleptic properties of the cheese. Therefore, in many cases, factors such as the type and/or amount of bacterial cultures, the production conditions and the ripening conditions which are used for the production of said cheese need to be varied in order to produce a healthy cheese which is accepted by the consumer in terms of its organoleptic properties. Another aspect of healthiness of the cheese lies in the presence of functional ingredients such as functional peptides, vitamins, nutraceuticals and the like.

Finally, a number of bacterial issues come together in quality, diversity and food safety of the cheese. Nowadays, most of the milk is pasteurised before the cheese making process starts. Such thermal treatment of the milk is preferred from a food safety point of view, since it reduces the presence of pathogenic micro-organisms in the cheese. However, the pasteurisation also results in the inactivation of many harmless microorganisms in the milk, which might otherwise positively influence the ripening process

and thereby the quality of the cheese. For these reasons, despite the potential hygiene risks, raw-milk cheese is still produced for a small but significant niche market of mainly certain artisanal and/or AOC- (i.e. "term of controlled origin") cheeses. At the same time, the cheese industry is looking for ways to safely employ raw milk in bulk processes for cheese making.

In view of the importance of cheese in everyday life, many efforts relating to research, development and quality control are spent in the field of cheese making. Thus, a lot of cheese is produced around the world, just for the purposes of - controlling the quality of the cheese, as largely determined by the starter cultures and the cheese making process; developing new variants of bacterial cultures, preferably starter cultures and adjunct cultures, by screening libraries of mutant bacterial cultures for mutants with altered activity; - developing new variants of existing cheese types, involving the selection of one or more bacterial culture(s) or functional ingredients such as enzymes, usually in combination with one or more modifications of the cheese making process; preventing or inhibiting the outgrowth of unwanted micro-organisms, including pathogens, in cheese, involving experimental work to modify the steps of the cheese production process and/or to develop bacterial cultures which prevent or inhibit the growth of said unwanted micro-organisms.

In view of the vast number of combinations of starter cultures, adjunct cultures, functional ingredients, including enzymes, and cheese production conditions which may be tested, the cheese making and screening exercises involved in the experimental work which is required for the above-mentioned purposes obviously involve a lot of resources, in particular time and raw materials.

As to raw materials consumption, cheese production in general requires a lot of milk - for example, approx. 100-120 litres of whole cow's milk are needed to produce one Gouda wheel of 12 kg. In view of this, protocols for making small cheeses have been developed to reduce raw materials consumption involved in activities pertaining to quality control, research and development of cheese and bacterial cultures.

For example, Shakeel-Ur-Rehman et al. (Lait 1998, 78, 607-620) describe a procedure for the small-scale manufacture of Cheddar-type cheese, employing 200 ml of milk per cheese. Although the procedure according to Shakeel-Ur-Rehman et al. - hereinafter further referred to as the Lait -procedure - provides for an appreciable reduction in milk consumption required for the production of a cheese for experimental purposes, as compared with the milk consumption required for a "normal" cheese as commercially marketed, in effect, the Lait -procedure still involves a significant volume of milk - i.e. 200 ml - per cheese which is produced.

Moreover, in terms of output capacity, the Lait-procedure allows for the manufacture of only 2 batches, i.e. 12 cheeses, in a working day.

SUMMARY OF THE INVENTION In view of the increasing need of the cheese industry to expand its capacity to test a vast and growing number of different bacterial cultures, proteolytic enzyme preparations and functional ingredients such as lipolytic enzymes, using many different cheese making and ripening conditions, there exists a need for a cheese making method which allows for a further reduced consumption of raw materials, in particular milk.

In addition, there exists a need for a cheese making method which allows for the production of a significantly greater number of cheeses for testing purposes per experimentalist per time unit, e.g. an 8h working day, as compared to the productivity of the cheese making methods hitherto known.

Moreover, there exists a need for a cheese screening method which allows for the screening of an increased number of cheeses, per experimentalist per time unit, for one or more features which are characteristic of, or which are determining for one or more organoleptic properties, health properties and/or food safety aspects of said cheeses.

Surprisingly, it has been found that these needs are met by a method for producing cheese in at least two cheese vats, the cheese vats being formed as an array of cheese

vats in a substrate, and each of the cheese vats having a volume ranging between 0.1 and 100 ml, the method comprising the sequential steps of: a. renneting in said cheese vats a cheese milk composition, the cheese milk composition comprising a volume of milk, at least one coagulant and at least one source of proteolytic enzymes to obtain curds; b. cutting the curds; c. optionally washing the curds; d. removing excess liquid from the curds; e. optionally drying the curds; and f. ripening the curds to obtain cheeses.

In one embodiment the cheeses obtained in step (f.) are in the same vats corresponding to the cheese vats wherein in step (a.) cheese milk composition were renneted. The volume of the cheese milk composition employed in step (a.) for the production of each cheese ranges between 0.1 and 100 ml and further preferably, said volume is equal to or smaller than the volume of the cheese vat, i.e. does not exceed the volume of the cheese vat, in which the cheese is produced.

Further preferably, the cheese milk composition corresponds by composition to the mixture comprising milk, one or more coagulants and one or more sources of proteolytic enzymes which are to be used in the renneting step for producing the corresponding cheese on a commercial scale. For example, if the method according to the present invention is to be used for producing Gouda-type cheese, the cheese milk composition is chosen within those ranges which are typical for a mixture comprising milk, one or more milk clotting enzymes and one or more bacterial cultures as common to the Gouda cheese-making protocol.

Advantageously with the present method cheeses have been produced using smaller volumes of milk than hitherto known. According to the method of the present invention, the volume of milk required for preparing one cheese is reduced by at least a factor of two, as compared with the Lait -procedure. Nonetheless, the amount of cheese produced appears still amenable for the analysis of at least one of its taste- and/or

flavour characteristics, with sufficient sensitivity, by techniques which are commonly employed for screening one or more features which are characteristic of, or which are determining for one or more organoleptic properties, health properties and/or food safety aspects of cheese, such techniques comprising GC and/or GC-MS, HPLC-MS, activity assays of enzymes involved in cheese ripening, promoter activity assays, and texture analysis.

In addition, it was found that the present method advantageously provides for performing at least one of the steps using equipment which recognizes the positions of the cheese vats within the substrate in which they are formed as an array, allowing for fast and automated processing of said at least one of the steps. Said equipment preferably comprises, for example, commercially available automatic dosing systems, e.g. for automatic filling of each of the cheese vats in step (a.), and/or automated equipment for cutting the curds in each of the cheese vats according to step (b.). Even more preferably, said equipment allows for fast, automated and parallel processing of at least one of the steps according to the method of the present invention. Parallel processing means that each required functionality can be simultaneously and independently performed on at least two cheese vats. For optimal compatibility with such automated equipment capable of recognizing the positions of the cheese vats in their substrate, in a preferred embodiment said substrate is a microtitre plate. In one embodiment the microtitre plate is a commercially available microtitre plate. In one embodiment the microtitre plate is a commercially available 12-wells, or 24-wells microtitre plate. In one embodiment the microtitre plate is a commercially available 48- wells microtitre plate. In one embodiment the microtitre plate is a commercially available 96-wells microtitre plate. In one embodiment the microtitre plate is a commercially available 384-wells, or 1536 wells microtitre plate. In one embodiment the volume of each of the cheese vats, in particular the wells of a microtitre plate, ranges between 0.1 and 100 ml, more preferably between 0.1 and 10 ml, more preferably between 0.1-2 ml, and more preferably between 1-2 ml. In one embodiment the microtitre plate is a 48-wells microtitre plate, each of the wells having a volume between 1 and 10 ml. In one embodiment, the microtitre plate is a 96-wells microtitre plate, each of the wells having a volume between 1 and 2 ml. In one embodiment the microtitre plate is a commercially available filter plate.

For example, if each of the cheese vats is provided by each of the wells of a 96-wells microtitre plate, in step (a.), the automated and parallel addition of the cheese milk composition to each of the 96 cheese vats requires an automated dispensing device carrying at least 2, or 4, or 8 or 12, or a plural of 8 or 12 up to 48 or 96 pipettes which can operate independently from each other and which can simultaneously fill at least 2, or 4, or 8 or 12, or a plural of 8 or 12 up to 48 or 96 of each of the 96 cheese vats, respectively. In one embodiment said automated dispensing device is commercially available as a 8 or 12 multichannelpipette. In one embodiment, depending on the variability of the cheese milk compositions within the plates, and typically employing between 6-12 microtitre plates per experimentalist per 8h day, a typical productivity could be obtained of between 576 - 1152 cheeses per experimentalist per 8h day. Compared to the Lait-procedure, according to this embodiment, the consumption of milk has been reduced by approximately two orders of magnitude, whilst the output capacity (in terms of number of cheeses produced per experimentalist per 8h day, using one set of equipment) has been increased by approximately up to two orders of magnitude.

More advantageously, the screening of the increased number of cheeses can be performed using techniques which lend themselves for high throughput screening. In addition, the cheese and/or the headspace over the cheese produced according to the method of the present invention could be successfully screened for one or more features which are characteristic of, or which are determining for one or more organoleptic properties, health properties and/or food safety aspects of said cheese. For example, as shown in the examples, cheeses produced using a 96-wells microtitre plate could be accurately analysed for their flavour profile using commercially available automated high throughput screening equipment, which recognizes the positions of the wells within the microtitre plate.

In one embodiment, individually prepared cheese is provided, the cheese having a volume of 0.01-15, preferably 0.1-10, more preferably 0.1-5, even more preferably 0.1- 1 ml, and the cheese being obtainable by the method according to the present invention.

DETAILED DESCRIPTION OF THE INVENTION

Each of the steps according to the method of the present invention will now be further illustrated below.

Step (a). 1. Further details of the ingredients comprised by the cheese milk composition.

The volume of milk comprised by the cheese milk composition may be obtained from any mammal, preferably from a cow, a goat and/or a sheep. As most commercially available cheese is still produced using whole milk, in one embodiment the volume of milk essentially contains whole milk. As the production of cheese having a lower fat content becomes increasingly important in view of healthiness, in another embodiment the volume of milk essentially contains partially defatted milk. Most bulk cheese is produced using pasteurised milk. Hence in one embodiment, the volume of milk contains pasteurised milk. Many artisanal and/or AOC-cheeses are produced using raw milk wherein for example control of outgrowth of unwanted micro-organisms is an important issue. This in another embodiment, the volume of milk preferably contains raw milk. In a further embodiment, the volume of milk contains a mixture of raw milk and pasteurised milk. It may be desired to balance aspects of organoleptic properties and food safety by employing a mixture of pasteurised milk and raw milk.

The at least one coagulant is preferably a milk clotting enzyme, such as calf rennet (e.g. Kalase, ex CSK Food Enrichment, The Netherlands), microbial rennet (e.g. Milase, ex CSK Food Enrichment, The Netherlands), chymosin (e.g., Chy-Max, ex Chr. Hansen A/S, Denmark), or any other suitable milk clotting enzyme. Most traditionally ripened cheese, such as Gouda, is produced using a milk clotting enzyme. In one embodiment, the coagulant is substantially free of milk clotting enzymes; in this case, the coagulant is preferably an acid, even more preferably a food grade acid, for example citric acid. The use of an acid as a coagulant is particularly preferred in the production of imitation cheese, such as pizza cheese.

Suitably the at least one bacterial culture effectively provides for at least one source of proteolytic enzymes. Most traditionally ripened cheeses are produced using at least one

bacterial culture as at least one source of proteolytic enzymes. However, it is also possible to fully or partially replace the at least one bacterial culture with a proteolytic enzyme preparation, which is preferably commercially available.

Thus the at least one source of proteolytic enzymes is preferably provided by at least one bacterial culture. In one embodiment, the bacterial culture comprises at least one starter culture. In another embodiment, the bacterial culture comprises at least one adjunct culture. In a particularly preferred embodiment, the bacterial culture comprises a mixture of at least one starter culture and at least one adjunct culture. Typically, the starter culture is employed for acidification of the milk and to affect the basic organoleptic properties of the cheese during ripening. The prevailing role of the adjunct culture is to further modify the taste, the flavour and/or the consistency of the cheese. In one embodiment, the bacterial culture comprises at least one new variant of a bacterium suitable for cheese making. Said bacterium may have been obtained by screening libraries of mutant bacterial cultures for mutants with altered activity. Furthermore, said bacterium may be used as a starter culture or as an adjunct culture.

Any embodiment providing for at least one bacterial culture as at least one source of proteolytic enzymes is in one embodiment suitable for providing a cheese milk composition which allows for the production of traditionally ripened cheese, such as Gouda-type cheese. In another embodiment, the at least one source of proteolytic enzymes is provided by a proteolytic enzyme preparation, suitably a proteinase and/or a peptidase. Both the proteinase and the peptidase are preferably derived from microbial sources, the proteinase is preferably derived from Bacillus, Aspergillus or Rhizomucor spp., and the peptidase is preferably derived from Aspergillus, Rhizomucor or Lactococcus lactis. According to this embodiment, a cheese milk composition is provided which is particularly suitable for the production of enzyme-modified cheese. In another embodiment, the at least one source of proteolytic enzymes is provided by a mixture of at least one bacterial culture and at least one proteolytic enzyme preparation. According to this embodiment, a cheese milk composition is provided which is suitable for the production of a large number of variants of traditionally ripened cheese, and/or for the acceleration of the ripening process of traditionally ripened cheese.

In a particularly preferred embodiment, the cheese milk composition further comprises at least one functional ingredient. In one embodiment a functional ingredient is an enzyme, more in particular a lipolytic enzyme, for example a lipase derived from animal or microbial sources. In another embodiment, if the at least one source of proteolytic ingredients does not contain any bacterial cultures, a functional ingredient may be an acid, preferably a food grade acid. If bacterial cultures are absent, and particularly if the volume of milk contains pasteurised or sterilised milk which is commonly the case for such cheese- making processes on a commercial scale, acidification of the cheese milk composition may be slow or even absent. In that case, an acid is preferably added to the cheese milk composition to lower the pH.

In other embodiments the functional ingredients may be protein(s) or peptide(s) with a beneficial functionality, for example protein or peptides produced by micro-organisms, or vitamins or nutraceuticals in general, such as vitamins, antioxidants, immunoactive peptides, and ACE inhibiting peptides.

In one embodiment, the one or more bacterial cultures comprises a strain that expresses reporter genes which are detectable in milk e.g. derivates of L. lactis MG5267 containing plasmid pNZ5519 which expresses the luciferase genes luxAB from Vibrio Harvey (Bachmann et al. Appl Environ Microbiol. 2007 Jul;73(14):4704-6). This is a genetically modified strain of particular interest for screening of, e.g., promotor- luciferase fusion constructs to assess promotor activity in cheese.

For purposes of evaluating food safety, in one embodiment, one or more bacteriocin- producing bacterial cultures and/or one or more unwanted micro-organisms are additionally comprised by the cheese milk composition. In one embodiment, the one or more bacteriocin-producing bacterial cultures and the one or more unwanted microorganisms are preferably added to the cheese milk composition. In one embodiment, the one or more unwanted micro-organisms are naturally present in the milk. In one embodiment, the one or more unwanted micro-organisms are added to the cheese milk composition. The one or more unwanted micro-organisms are preferably pathogenic and/or spoilage micro-organisms. A preferred pathogenic micro-organism is a strain of

the species Listeria, Enter ococcus, Staphylococcus, Streptococcus or Bacillus or moulds or fungi. A preferred spoilage micro-organism is a species of the Clostridium strain.

Step (a.) 2. Dosing of the ingredients.

For ease of addition and compatibility with automatic dosing equipment, the at least one coagulant and the at least one source of proteolytic enzymes are dosed in the form of an aqueous solution and/or suspension. Preferably, the cheese milk composition does not contain more added water than needed for dosing the other ingredients for making up the cheese milk composition, in order not to excessively dilute the milk solids. However, the at least one coagulant and the at least one source of proteolytic enzymes should not be dosed in a too concentrated form, since practical problems of viscosity, stability and/or dosage accuracy could occur. It turns out that for these reasons, and in order to match commercially relevant cheese making processes as closely as possible, the at least one coagulant is preferably dosed in the form of a concentrated aqueous solution, said concentrated aqueous solution preferably having a strength which is commercially useful. If the at least one coagulant is provided by at least one milk clotting enzyme, the at least one milk clotting enzyme is preferably employed at a strength of between 100-1000 IMCU per ml of concentrate. Further, if the at least one source of proteolytic enzymes comprises at least one bacterial culture, the at least one bacterial culture is preferably dosed in the form of a concentrated aqueous suspension, the number of bacteria in said concentrated aqueous suspension preferably ranging between 1OE ⁷ and 1OE ¹² cfu/ml, more preferably ranging between 10 ⁸ -10 ^π cfu/ml.

In practice, this means that a volume of at least 80%, preferably at least 90% or at least 95%, and at most substantially all or at most 99.99% or at most 99% or at most 95% of the cheese milk composition is provided by the volume of milk. In one embodiment, if the at least one coagulant is provided by at least one milk clotting enzyme, the activity of the at least one milk clotting enzyme in the cheese milk composition ranges between 5 and 150 IMCU, more preferably between 10 and 50 IMCU, per litre of cheese milk composition. In another embodiment, if the at least one source of proteolytic enzymes comprises at least one bacterial culture, the cheese milk composition preferably

comprises between 10 ⁴ and 10 ¹⁰ cfu/ml, more preferably between 10 ⁶ -10 ⁸ cfu/ml, of bacteria originating from said at least one bacterial culture.

In one embodiment, the cheese milk composition is prepared by dosing the required amounts of ingredients into the cheese vat - if necessary, followed by mixing. In another embodiment, the cheese milk composition is fully or partially prepared as a pre-mix, using a certain amount of milk and one or more of the other ingredients as required to prepare a composition which equals the cheese milk composition to be dosed into the cheese vats according to step (a.). The volume of milk employed in preparing the pre-mix in one embodiment ranges between 20 and 200 ml, so that said other ingredients can be even more accurately dosed in the required amounts.

In one embodiment, the cheese milk composition is partially prepared as a pre-mix. Preferably, the pre-mix is made up of a volume of milk (preferably having a temperature which is slightly below the renneting temperature), comprising the coagulant and one or more starter cultures (as part of the at least one source of proteolytic enzymes). The appropriate volume of the pre-mix is dosed into each of the cheese vats, and then, one or more adjunct cultures are individually added to the contents of the cheese vats to make up for the cheese milk composition. The cheese vats are then inoculated at the renneting temperature to allow step (a.) to proceed. Using automated pipetting equipment, the filling of the cheese vats using the premix and the subsequent dosing of the one or more adjunct cultures is done within 2 minutes, so that step (a.) is allowed to commence for each well at virtually the same time.

In one embodiment the total volume of the cheese milk composition preferably ranges between 20 and 90% of the volume of the cheese vat in which the cheese is produced.

Step b. For performing the cutting step (b.), in one embodiment, the coagulum, or curds, obtained in step (a.) is cut using cutters made of wires stretched a certain distance apart across a frame. According to this embodiment, the volume of the cheese vat preferably ranges between 5-100 ml, and the distance between the wires preferably ranges from a few tenths of a cm to 1 cm, depending on the volume.

In another embodiment, wherein the volume of the cheese vat is smaller than 5 ml, specialised equipment is preferably used for cutting the curds. A preferred device for cutting the curds in a cheese vat having a small volume comprises a pin-shaped bar which is allowed to penetrate the contents of the cheese vat in a vertical direction, following which the bar is manually or automatically moved in a series of directions within a horizontal plane, in order to effectively cut up the curds in pieces, to provide a two-phase system comprising curds particles and liquid.

According to one embodiment, a special device has been developed for cutting the curds which allows for the simultaneous stirring of the contents of each of the cheese vats provided by each of the wells of a 96-wells microtitre plate, in particular each of the wells having a volume of between 1-2 ml. Said device is equipped with one pin- shaped bar per cheese vat. In performing the cutting step (b.), each pin-shaped bar is simultaneously allowed to penetrate the contents of the cheese vat in a vertical direction, following which each bar is manually or automatically moved in a series of directions within a horizontal plane, in order to effectively cut up the curds in pieces within each cheese vat. Correspondingly, for simultaneously cutting the curds contained in each of the 96 cheese vats of dimensions 7.3x7.3x4 lmm as provided by a Greiner Bio-One master block 96 well microtitre plate (#78027x), a device has been developed comprising 96 pin-shaped bars, each of dimensions 1x50 mm. Satisfying results were also obtained using a stirring device made from stainless steel and consisting of a plate with a handle at the top and 96 pin-shaped bars attached to the bottom. The pin-shaped bars are 3 mm in diameter and 45 mm long, and they are aligned in a way so they are positioned precisely in the middle of each well if the stirring device is placed in a 96-well microplate. Cutting of the curds is preferably performed by slow horizontal and vertical movements of this device through the curds. For providing a robust setup for effectively cutting of the curds, the bottom side of the pin-shaped bars are preferably moved within a distance of between 100-3000 micrometers from the bottom of each cheese vat. Further preferably, each of the pin- shaped bars is forced to produce a rotary movement in the horizontal plane at a speed which preferably ranges between 0.3-30 rotations per second., and the rotary movement is performed for a time of between 2-30 min, depending on the on the protocol for preparing a certain desired type of cheese, hereinafter also referred to as cheese protocol.

In a preferred embodiment, the cutting step (b.) comprises the production of curds particles by agitation of the curds using a pin-shaped device, e.g. the pin-shaped device as described above, until a desired average particle size is obtained, the desired average particle size preferably ranging between 0.1-0.9 mm, preferably 0.2-0.4 mm for Gouda- type cheese, or 0.4-1.2 mm, preferably 0.8-1.0 mm for Cheddar-type cheese.

In one embodiment, step (b.) is optional.

Step c. Depending on the cheese protocol, a washing step may be performed after cutting the curds. For example, a Cheddar protocol usually does not employ a washing step, but the inclusion of washing step (c.) is mandatory for preparing cheese according to a Gouda protocol. In the Gouda protocol, the function of the washing step is essentially to remove some of the lactose present in the whey. In addition, the washing liquor may contain a preservative, such as sodium nitrate, to protect the cheese during ripening.

Accordingly, the washing step replaces a certain volume of a liquid comprising lactose by a washing liquor, the washing liquor being an aqueous solution which is preferably lactose-free, and which may further contain an additive, such as a preservative.

Correspondingly, the washing step (c.) according to the present preferably comprises the sequential steps of i. centrifuging the composition obtained after cutting step (b.); ii. removing the supernatant from the curds; iii. re-suspension of the curds in an aqueous solution.

The composition of the aqueous solution employed in sub-step (iii.) preferably corresponds to the washing liquor which is used in a process for preparing the corresponding cheese on a commercial scale.

Sub-step (i.) in combination with sub-step (ii.), i.e. centrifugation followed by removal of supernatant from the curds, is highly advantageous in order to allow a proper

separation of the supernatant from the curds within a reasonable time span. Without being bound to theory, it is thought that the benefit for employing centrifugation as part of the washing step is associated with the relatively small average particle size of the curds obtained in step (b.) - the small average particle size being imparted by the cheese vats having a small volume. For example, the average size of the curds particles obtained in a commercial-scale cheese production process according to a Gouda protocol is approximately 0.5-1 cm whilst for obtaining Gouda-type cheese according to the method of the present invention, using cheese vats having a volume of between 1-2 ml, the preferred average size of the particle size of the curds obtained in step (b.) typically ranges between 0.2 and 0.4 mm.

Further preferably, the centrifugation is performed at a centrifugal force of between 100-1500 g, more preferably of between 800-1000 g. Further preferably, the centrifugation is performed within a time span of between 30 seconds and 1 hour, more preferably of between 1-10 mins.

It appeared that these parameters provide the best results for washing the curds particles obtained in step (b.). If the centrifugal force and/or the duration of the time span for performing the centrifugation is smaller than the lowest values of the ranges indicated above, the separation of curds and supernatant is less efficient and/or complete. If the centrifugal force and/or the duration of the time span for performing the centrifugation is higher than the highest values of the ranges indicated above, the curds will be compacted to such an extent as to hamper the re-suspension of the curds particles in an aqueous solution.

Step d.

This step concerns the removal of excess liquid from the curds. One of the important functions of this step is to regulate the moisture content of the curds before ripening. In commercial-scale processes for the manufacture of cheese, the corresponding step is usually performed by pressing. It appeared that according to the method of the present invention, the removal of excess liquid from the curds advantageously is carried out by centrifugation of the - optionally re-suspended - curds, followed by removal of supernatant from the curds.

In a preferred embodiment, liquid removal step (d.) comprises subjecting the, optionally re-suspended, curds to a centrifugal force of between 2000-5000 g, preferably of between 2500 and 3500 g, during a time span of between 0.5-5 hours; and - removing the supernatant from the curds.

If the centrifugal force and/or the duration of the time span for performing the centrifugation is smaller than the lowest values indicated in the ranges above, not enough supernatant can be removed from the curds, so that the moisture content of the curds before ripening is excessively high. If the centrifugal force is higher than 500Og, most commercially available mictrotitre plates start to break down under the force. For example, a Greiner Bio-One master block 96 well microtitre plate (#78027x) is specified to withstand 4800 g. If the cheese vats - especially the microtitre plates - are strong enough, higher centrifugal forces may be employed, for example up to 700Og, lOOOOg or 2000Og. If the duration of the time span for performing the centrifugation is longer than 5h, the extra centrifugation time hardly provides for a further amount of supernatant which can be removed.

In one embodiment, it may be advantageous that the liquid removal step is applied using a so-called filterplate. This is a 96 well microtitre plate having a volume of between 1-2 ml, or a 48 well having a volume of approx. 5 ml, wherein each of the wells is equipped with a filter at the bottom, allowing the liquid (which otherwise, without the use of the filter, is called supernatant) to escape through the bottom of the well during the centrifugation of the (optionally resuspended) curds. Such filterplates are for example commercially available, e.g. from Porvair and Millipore.

Without being bound to theory, it is believed that the use of a centrifugal liquid removal step, possible in conjunction with the relatively small size of the cut cheese curds obtained in step (b.), may result in a fat loss during the cheese making process of the present invention, especially during the liquid removal step (d.), which fat loss may be higher than compared with a conventional cheese making process, especially when compared with a conventional industrial scale cheese making process. Nevertheless, such an additional fat loss would not be detrimental for the microstructure of the cheese

obtained using the method of the present invention, wherein the liquid removal step (d.) comprises a centrifugation step. Furthermore, such an additional fat loss would not noticeably affect proteolysis, flavour generation or any other cheese property that is determined within the framework of the present invention.

In cases where it is desired to compensate for said extra fat loss that would occur, a predetermined amount of milk fat may be added to the cheese milk composition in step (a.). The predetermined amount of milk fat may be conveniently added as cream. Alternatively, the extra fat loss may be reduced by employing a microtitre plate wherein the exposed surfaces of each of the wells is provided with either hydrophilic surfaces or surfaces which have a low surface energy. A hydrophilic surface is preferably a glass surface. A surface having a low surface energy is preferably selected from the group consisting of a polyethylene, a polypropylene, or a halogen containing plastic such as Teflon® or PVDC. The extra fat loss may be further reduced by employing a mictrotitre plate equipped with a filter mat (see above); it is preferred that the filter mat is made of or coated with a hydrophilic material, such as glass, or a material having a low surface energy, such as a polyethylene, a polpropylene, or a halogen containing polymer such as Teflon or PVDC. Finally, in the liquid removal step, the centrifugation step may be replaced by a pressing step, as in conventional cheese making; this pressing step comprises submitting the cut curds obtained in step (b.) or the optionally washed cheese curds obtained in step (c) to a compressive force whilst allowing excess moisture to escape. The compressive force is preferably homogeneously applied to the cheese curds and is (hence) preferably non-centrifugal.

In a preferred embodiment, liquid removal step (d.) comprises subjecting the, optionally re-suspended, curds to a compressive force of between 1 and 1000 N, preferably of between 5 and 100 g, during a time span of between 0.5-5 hours; and allowing excess liquid to escape

The pressing step may comprise transfering the cut curds obtained in step (b.) or the optionally washed curds obtained in step (c.) on a filtermat which is comprised by another microtitre plate. Herein, the compressive force is preferably applied using a

stamp having a diameter which fits the bore of the wells. The excess liquid is conveniently allowed to escape through the pores of the filtermat, and thus removed from the cheese curds.

Alternatively, the compressive force is applied using a porous stamp, and the excess liquid is allowed to escape through the porous stamp. Preferably, the excess liquid is then removed, preferably by suction or decanting.

Brining step

Commercially produced cheese almost always contains a certain amount of added sodium chloride. The sodium chloride is commonly added during the commercial cheese making process, using a brining step. For this reason, in one embodiment, the method according to the present invention further comprises a brining step. In order not to negatively interfere with the acidification of the milk, as determined by the at least one bacterial cultures, the brining step is performed at any moment after cutting step (b.).

In order to be able to accurately control the amount of sodium chloride which enters the curds, in a preferred embodiment, the brining step is performed after liquid removal step (d.). Moreover, since the presence of sodium chloride is of influence for the development of one or more of the organoleptic properties of the cheese, in a further preferred embodiment, the brining step is performed before the ripening step (f).

Disappointingly, it turned out that the brining step according to the Lait-procedure, when employed in the production of semi-hard and/or hard-type cheese (such as Gouda, Cheddar, etc.) could not be used in the method according to the present invention. The Lait-procedure teaches the soaking of a centrifuged mass of curds of approx. 20 grams in a brine bath (containing approx. 20% NaCl by weight) for 30 mins. However, it was found that this procedure produced unacceptably high salt levels within the curds produced according to the method of the present invention. Although shorter soaking times resulted in lower salt levels, the NaCl content in the curds after brining was still not sufficiently reproducible.

Surprisingly, it was found that the addition to curds of a controlled amount of sodium chloride is an effective way to produce cheeses having a reproducible salt content. The

sodium chloride is preferably added as an aqueous solution. Furthermore, in one preferred embodiment, the brining step comprises the addition of a controlled amount of an aqueous solution comprising sodium chloride to the curds, the volume amount of the aqueous solution comprising sodium chloride preferably ranging between 0.1 and 2 vol.% with respect to the volume of the milk employed in step (a.), and the known concentration of sodium chloride preferably ranging between 50 - 300 g/L. Within these ranges, the volume of the aqueous solution comprising the sodium chloride is high enough to be accurately measured and dosed into each of the cheese vats holding the curds.

It is noted that, within the volume and concentration ranges indicated above, it has been observed that the solution comprising the sodium chloride, after dosing into the cheese vat, rapidly penetrates the curds. Hereby, the desired sodium chloride concentration is reached in the curds, and at the same time, the moisture content of the curds is increased.

In a particularly preferred embodiment, wherein the cheese vats have a volume smaller than 5 ml, the cheese vats are briefly centrifuged after addition of the aqueous solution comprising the sodium chloride, in order to make sure that the volume of said aqueous solution will be entirely taken up by the curds. Without briefly applying the centrifugal force, if drops of the aqueous solution comprising sodium chloride would accidentally hit the walls of the cheese vat when dosed into each of the cheese vats holding the curds, a certain amount of said aqueous solution could later dry up against the walls of the cheese vats instead of contributing to the salt content of the curds.

According to a further preferred embodiment, the brining step is performed after liquid removal step (d.) and before the ripening step (f). Correspondingly, provided that the volume amount of the aqueous solution comprising sodium chloride which is dosed to the curds preferably ranges between 0.1 and 2 vol.% with respect to the volume of the milk employed in step (a.), and provided that the concentration of sodium chloride in said aqueous solution ranges between 50 - 300 g/L, the brining step will preferably not increase the moisture content of the curds by more than 10 percent points, whilst still providing the curds with the desired amount of sodium chloride.

In a further preferred embodiment, for achieving a certain desired concentration of sodium chloride in the cheese before the start of the ripening step, the total amount of sodium chloride to be added to the cheese vat is calculated using e.g. the Van Slyke and Price Formula. For a review on how to calculate the yield of cheese see the review by Emmons et al. in J. Dairy Sci. 1990, 73: 1365-1394. The desired concentration of sodium chloride in the cheese after the brining step preferably ranges between 1-5% by weight, on dry matter of the cheese.

In one embodiment, the aqueous solution of sodium chloride comprises one or more unwanted bacteria which are commonly found in commercial brine baths. This embodiment favorably allows for the production of cheese under commercially relevant brining conditions, and for the subsequent screening of the cheese thus produced for one or more features which are characteristic of, or which are determining for one or more food safety aspects of the cheese.

Step (e.). In attempts to produce cheese of the semi- hard or hard type (such as Cheddar or Gouda-type cheese) using the present method, it appeared that the moisture content of the curds obtained after step (d.) was significantly higher than expected. It appeared that in using the centrifugation method according to the Lait -procedure for removing excess liquid from the curds according to step (d.), the moisture level of the curds could not be reduced to satisfactory levels. For example, centrifugation method according to the Lait -procedure when applied to suspended curds particles which had been obtained using a Gouda-type protocol, yielded a moisture content of the curds which could not be reduced to below 50% (w/w), even when the highest g-forces tolerated by a commercially available plastic 96-wells microtitre plate were applied for an appreciable time, e.g. 30 mins.

Given the known importance of the moisture content on the ripening stage, particularly on the kinetics of enzymatic conversions which are relevant for the development of taste and texture, it follows that the moisture content of the cheeses produced following step (d.) according to Lait -procedure is in most cases too high to be representative for a commercially produced cheese of the semi-hard or hard-type, containing the same ingredients.

If a brining step is employed, the moisture issue becomes even worse, as the brining step further provides some additional moisture to the curds, for example, typically, around 5%.

To illustrate the problem, the moisture content of a Gouda-type curds obtained after step (d.) typically ranges 50-55%; following the brining step, the moisture content of the same curds has typically reached between 55-60%. By comparison, the target for the moisture content of Gouda-type curds before ripening is approximately 35-45%.

To solve the problem of the curds for the production of semi-hard and hard cheeses having a too high moisture content prior to ripening, the curds have to be further dried prior to the ripening step.

It was found that a significant reduction of the moisture content of the curds could be obtained by allowing moisture from the curds to evaporate in an atmosphere of controlled temperature and relative humidity (RH), wherein the RH < 100%. Preferably, said atmosphere is controlled to a relative humidity of lower than 80%, more preferably of lower than 70%. Further preferably, said atmosphere is controlled to a relative humidity of 0%, more preferably at least 10%, even more preferably at least 20%.

Further preferably, said atmosphere is controlled to a temperature between 0 and 30 ⁰ C, even more preferably between 5 and 25 ⁰ C, most preferably between 10 and 20 ⁰ C. It is especially preferred that said atmosphere is anaerobic and/or sterile. It is most preferred that said atmosphere is anaerobic and sterile. A sterile and anaerobic atmosphere may be conveniently provided in a closable vessel, for example a jar. According to a preferred embodiment, said vessel is preferably sterilised and filled with an inert gas, after which the curds are placed in the vessel to undergo the drying step. After placing the curds in the vessel, the vessel is preferably closed in order to keep its contents sterile and anaerobic. The vessel is preferably provided with a moisture controlling agent, for example silica gel, difosforous pentoxide, or a saturated salt solution, in order to being able to control the rate of evaporation of the moisture from the curds. It appeared that performing the drying step at a too high relative humidity results in the drying time being excessively long. Drying times longer than 48h are to be avoided

according to one preferred embodiment of the present invention, in order to prevent the ripening of the curds from occurring already during the drying step (e.) to a noticeable extent. For this reason, in one embodiment, the drying step (e.) is preferably performed within a time span of less than 48 hours, more preferably within a time span of less than 24 hours.

However, if the relative humidity is too low, the drying takes place too rapidly, so that the moisture content of the curds before ripening cannot be accurately controlled. Likewise, if the temperature is too high, ripening of the cheese will occur too readily during the drying step (e.), which needs to be avoided. If the temperature is too low, the drying process will be too slow in order to be practically useful.

Optimal settings for the drying step (e.) were obtained if the controlled atmosphere was set at a relative humidity between 0 and 80%, more preferably between 10 and 70%, and to a temperature between 5 and 40 ⁰ C, more preferably between 10 and 25 ⁰ C. According to this embodiment, the moisture content of the cheeses could be carefully controlled by submitting the cheeses to the drying conditions for a certain amount of time, until the target value for the moisture content was reached. Within the preferred ranges of temperature and relative humidity according to this embodiment, and within a time span of less than 48 hours, preferably of less than 24 hours, the drying step favourably effectuates a carefully controlled reduction of the moisture content of the curds of between 5-35 percent points, by weight, whilst limiting the extent of ripening of the curds to occur during the drying step to the lowest possible extent.

In a further preferred embodiment, the drying step (e.) comprises the evaporation of moisture from the curds obtained after step (d.) through a membrane, wherein the membrane separates the headspace over the pressed curds from the atmosphere into which the moisture from the curds obtained after step (d.) is evaporated. Preferably, the pore size of the membrane is smaller than 0.2 μm in order to allow the membrane to protect the curds from any micro-organisms which could enter the cheese vat from the controlled atmosphere. It is further preferred that the water vapour permeability of the membrane is at least 300 in order not to significantly reduce the moisture evaporation kinetics. The present embodiment conveniently allows the controlled

atmosphere to be provided by a commercially available climate chamber, without the need for additional measures, for example to disinfect the air, within said chamber. Preferred membranes according to the present embodiment include: Breathseal Greiner Bio-One #676050 (water vapour permeability as specified: 4200 g.m ^"2 .24h ^"1 ); Breath- Easy® Diversified Biotech (water vapour permeability as specified: 700 g.m ^"2 .24h ^"1 ); Breath-Easier® Diversified Biotech (water vapour permeability as specified: 4200 g. m ^"2 .24h ^"1 ); Excel Scientific AreaSeal BlOO (water vapour permeability as specified: 4200 g. m ^"2 .24h ^"1 ).

Step f.

In one embodiment, the ripening temperature employed in step (f.) ranges between 0 and 40 ⁰ C, preferably between 5 and 30 ⁰ C, even more preferably between 10 and 25 ⁰ C. The narrow temperature range typically corresponds to values which are commonly employed during the ripening of cheeses on a commercial scale.

Further preferably, the ripening time ranges between 4 hours - 365 days, preferably between 1-4 weeks.

In a further preferred embodiment, prior to the ripening step (f), each of the cheese vats is closed off using a seal which is essentially impermeable to water vapour and/or gas, the gas preferably comprising N ₂ , O ₂ and/or CO ₂ . The use of a seal which is essentially impermeable to water vapour prevents evaporative water loss from the cheese during ripening. This is important, since the evaporation of a few tenths of a gram of water already leads to a significant drop of the moisture content of the ripening cheese.

In another preferred embodiment, the headspace of the curds has been replaced by an inert atmosphere prior to the drying step (e.) and/or prior to the ripening step (f). This embodiment is particularly favourable if the growth of aerobic micro-organisms in and/or on the cheese is to be excluded during ripening.

In one embodiment, prior to ripening step (f), the cheeses are provided with one or more cultures for surface ripening. According to this embodiment, if one or more of the

surface cultures are obligate aerobic, the headspace over the cheese is preferably not replaced by an inert atmosphere.

In a particularly preferred further embodiment, the ripening step (f.) is performed within the confinements of a cheese vat. Further preferably, the curds do not leave the cheese vat after production of the curds produced in step (a.). Further preferably, one or more organoleptic properties of the correspondingly produced cheese after ripening are analysed using the cheese and/or the headspace over the cheese, the analysis being performed within and/or from within the same cheese vat in which the ripening step (f.) was performed. Alternatively, in one embodiment, the cheeses are transferred from the cheese vat in which they were ripened to a container, e.g. a vial, within or from within the analysis is performed. Further preferably, said transfer of the cheese from the cheese vat into said vial, and/or said analysis is performed using equipment which recognizes the positions of the cheese vats within the substrate in which they are formed as an array, allowing for automation of said transfer and/or said analysis, respectively. Further preferably, the analysis is performed using equipment particularly comprising so-called high throughput screening techniques. Particularly preferred examples of high -throughput screening techniques for screening one or more features which are characteristic of, or which are determining for one or more organoleptic properties of the cheese produced according to the present invention include GC-MS, HPLC MS, capillary electrophoresis, MALDI-TOF, gene expression analysis, activity assays of enzymes in particular enzymes involved in cheese ripening, promoter activity assays, and texture analysis. Gene expression analysis is preferably performed as (Q)- PCR, using a RNA template, or as microbial population dynamics analysis, using a DNA template; herein, RNA or DNA microarrays, respectively, are preferably employed. Herein, for example, GC-MS is employed to screen the headspace over a cheese for the presence of volatile organic components, the ensemble of volatile organic components in the headspace being a feature which is characteristic for the flavour profile of the cheese. As another example, a proteolysis assay can be employed to determine proteolysis activity of a number of relevant enzymes in the cheese, the proteolysis activity being a feature which is characteristic for the taste profile of the cheese at the time of performing the analysis. At the same time, proteolysis activity is also a feature which is determining for the taste profile of the cheese, in that a

proteolysis activity, producing breakdown of casein during ripening, can be correlated to the development of taste during ripening.

Thus, the invention also concerns screening of the cheese prior to or during or after ripening for one or more features which are characteristic of, or which are determining for one or more organoleptic properties, health properties and/or food safety aspects of the cheese.

In one embodiment, a method is provided for analysing cheese obtained according to the method of the present invention, said method comprising a screening of the cheese and/or the headspace over the cheese for one or more features which are characteristic of, or which are determining for one or more organoleptic properties, health properties and/or food safety aspects of said cheese.

In one embodiment, a method is provided for analysing cheese obtained according to the method of the present invention for one or more flavour properties, wherein the at least one source of proteolytic enzymes comprises a starter culture and a plurality of different adjunct cultures, wherein the adjunct cultures are varied, and wherein the headspace over the cheese is screened for volatile organic components. In one embodiment, a method is provided for analysing cheese obtained according to the method of the present invention for one or more taste properties, wherein the at least one source of proteolytic enzymes comprises a starter culture and a plurality of different adjunct cultures, wherein the adjunct cultures are varied, and wherein the cheese is screened using one or more assays for the activity of enzymes involved in cheese ripening. In one embodiment, a method is provided for analysing cheese obtained according to the method of the present invention, wherein the at least one source of proteolytic enzyme further comprises proteolytic enzyme preparations and/or lipolytic enzymes, and wherein the proteolytic enzyme preparations and/or lipolytic enzymes are varied. In one embodiment, a method is provided for analysing cheese obtained according to the method of the present invention, wherein the cheese milk composition comprises one or more bacteriocin-producing bacterial cultures and/or one or more unwanted micro-organisms, wherein the bacteriocin-producing bacterial cultures and/or the unwanted micro-organisms are varied, and wherein the cheese is screened for the

presence of the one or more bacteriocins and/or one or more unwanted microorganisms.

In one embodiment, a method is provided for analysing cheese obtained according to the method of the present invention, wherein the cheese milk composition comprises one or more functional ingredients, wherein the functional ingredients and/or the at least one source of proteolytic enzymes are varied, and wherein the cheese is screened for the presence of health-promoting components.

In one embodiment, said health-promoting components comprise functional peptides including ACE inhibitors, satiety -inducing peptides and taste enhancing peptides. Peptide profiles and levels can be modulated by providing the cheese milk composition with varying amounts and types of starter cultures and/or adjunct cultures and/or preparations of proteolytic enzymes in the presence of precursor molecules in varying amounts and types.

In one embodiment, the cheese obtained according to the method of the present invention is screened for profiles and levels of nutraceuticals such as vitamins (including vitamin B2,B11, B 12, K) and antioxidants. Profiles and levels can be modulated by providing the cheese milk composition with varying amounts and types of starter cultures and/or adjunct cultures in the presence of precursor molecules and/or enzymes in varying amounts and types.

In one embodiment, the cheese obtained according to the method of the present invention is screened for the production of unsaturated fatty acids from fatty acids derived from milkfat or production de novo. Profiles and levels can be modulated by providing the cheese milk composition with varying amounts and types of starter cultures and/or or adjunct cultures producing fatty acid desaturases, and/or by providing the cheese milk composition with varying amounts and types of precursor molecules and/or enzymes.

Thus, in one embodiment, a method is provided for analysing cheese obtained according to the method of the present invention, wherein the cheese milk composition comprises one or more functional ingredients comprising a precursor molecule for a health-promoting component, wherein the precursor molecule for the health promoting

component and/or the at least one source of proteolytic enzymes are varied, and wherein the cheese is screened for the presence of the health-promoting component, said health-promoting component comprising one or more functional peptides including ACE inhibitors, satiety- inducing peptides and taste enhancing peptides.

Thus, in one embodiment, a method is provided for analysing cheese obtained according to the method of the present invention, wherein the cheese milk composition comprises one or more functional ingredients comprising a precursor molecule for a health-promoting component, wherein the precursor molecule for the health promoting component and/or the at least one source of proteolytic enzymes are varied, and wherein the cheese is screened for the presence of the health-promoting component, said health-promoting component comprising one or more nutraceuticals, preferably one or more vitamins and/or one or more anti-oxidants.

Thus, in one embodiment, a method is provided for analysing cheese obtained according to the method of the present invention, wherein the cheese milk composition comprises one or more functional ingredients comprising an enzyme or a precursor molecule for a health-promoting component and wherein the at least one source of proteinase is capable of producing fatty acid desaturases, wherein the enzyme and/or the precursor molecule for the health promoting component and/or the at least one source of proteolytic enzymes are varied, and wherein the cheese is screened for the presence of the health-promoting component, said health-promoting component comprising one or more unsaturated fatty acids.

In one embodiment, a method is provided for analysing cheese obtained according to the method of the present invention, wherein the ripening conditions are varied or further varied.

In one embodiment the invention further provides a method for analysing one or more features which are characteristic of, or which are determining for one or more organoleptic properties of the cheese produced according to the present invention, said method comprising analysing the cheese and/or the headspace over the cheese using one or more high-throughput screening techniques. This embodiment thus allows a

further increase in productivity, in that at least one of the steps in the production of, and at least one of the steps in the sequential screening of one or more of the organoleptic properties of the cheese are at least partly performed using automated equipment. In one embodiment, a method is provided for analysing cheese obtained according to the method of the present invention, the analysis involving the screening of the cheese prior to or during or after ripening for one or more features which are characteristic of, or which are determining for one or more organoleptic properties, health properties and/or food safety aspects of the cheese, the screening being performed using one or more high-throughput screening techniques, and the one or more high-throughput screening techniques preferably being selected from the group consisting of GC, GC- MS, HPLC MS, capillary electrophoresis, MALDI-TOF, gene expression analysis, activity assays of enzymes involved in cheese ripening, promoter activity assays, and texture analysis.

EXAMPLES

Example 1. Protocol for preparing Gouda-type cheese, using the method according to the present invention, and production for preparing Gouda-type cheese employing one starter culture and varying 2 adjunct cultures.

For the production of Gouda type cheese, standardized cow's milk with a fat content of 3.6% and a protein content of 2.5% was used. A commercially available starter culture (Bos, ex CSK Food Enrichment Ede, The Netherlands) was thawed and subsequently grown in a small volume of milk at 20 ⁰ C for 20 hours. 1 part of the Bos culture thus obtained was passaged into 100 parts of milk, and the culture was grown again at 20 ⁰ C overnight. The resulting culture is further referred to as the pre-incubated Bos starter culture.

Then, Kalase (150 IMCU/ml) (ex CSK Food Enrichment, Ede, The Netherlands), calcium chloride solution (33% wt/vol. in water) and the pre-incubated Bos starter culture were added, with brief mixing, to a volume of standardized cow's milk, for obtaining a final concentration of 1.2 mM CaCl ₂ , 0.5% Bos-starter culture, and 0.23 g Kalase per litre of standardized cow's milk.

Each of the 96 wells of a microtitre plate (Greiner Bio-One master block 96 well, #78027x) were then filled with the mixture obtained above, and the cheese milk composition was produced to completion by the addition to the contents of each individual well of 0.5% (v/v) of each of two different adjunct cultures, the adjunct cultures being selected from two different thermophilic lactobacilli -APS and T72 . Thus, 48 wells comprised the first adjunct culture and 48 wells comprised the second adjunct culture. In each case, the volume of the cheese milk composition thus obtained amounts to ca. 1.7 ml.

The microtitre plate was then sealed with a capmat to avoid evaporation, spillage and contamination of the cheese milk composition, and subsequently the cheese milk composition was incubated at 30.5 ⁰ C. After 30-45 min, the cheese milk composition had coagulated, and the curds thus formed were subsequently cut using a device comprising 96 pin-shaped bars, each of dimensions 1x50 mm. Cutting was performed by allowing each of the pin-shaped bars to penetrate the coagulum, following which each of the pin-shaped bars was forced to produce a rotary movement in the horizontal plane at a speed of approx. 1 rotation per second, for 10 mins. During the cutting step, the bottom side of the pin-shaped bars remained within a distance of between 100-3000 micrometers from the bottom of each cheese vat. After cutting the curds, the contents of the cheese vats were "tumbled" by sealing the plates with a capmat and placing the plate on a tumbler for 20 min. (at 30.5C) at about 20 rotations per minute in order to promote syneresis (i.e. separation of liquid from the curds).

After the cut curds were allowed to settle to the bottom for 2 min, the curds were washed. Accordingly, the microtitre plate was centrifuged at 800 g for 5 min in order to compact the curds, and subsequently 680 μl of whey were removed from each cheese vat. Then, 595 μl of sterile water (at 40 ⁰ C was added to each well, and the curds were resuspended by briefly stirring the contents of each cheese vat using the cutting device described above, followed by tumbling of the microtitre plate at 35.5 ⁰ C for 40 min at 20 rpm. Afterwards the plate was incubated at 35.5 ⁰ C for 20 min. without tumbling, to complete the washing step.

To separate the curds from the liquid, the micro titre plate was centrifuged at 3000 rpm (1865 g) for 60 min at 35.5 ⁰ C. The supernatant was removed by decanting the micro titre plate and placing it upside down on a sterile tissue for 30 min. The plate was then sealed using a capmat (Greiner #381081) and incubated at 25 ⁰ C overnight to allow acidification of the curds.

Twenty hours after separating the curds from the liquid 17 μl of a 20% NaCl solution wt/vol was added to the contents of each individual cheese vat, resulting in a salt content of approximately 3% NaCl (based on dry matter of the cheese). The required amount of sodium chloride to be added to each cheese vat was calculated using the Van Slyke and Price Formula, assuming that each cheese was produced from 1.7 ml of milk having a fat content and 3.6% and a casein content of 2.5%. To make sure that the NaCl solution quantitatively reached the curds and did not partially stick to the walls of the cheese vats, the plate was centrifuged at 1000 g for 1 min. Then the plate was then sealed using a gas and water permeable membrane which is impermeable to unwanted bacteria (Greiner #381081) and placed in a climate chamber set to 17 ⁰ C and 70% relative humidity in order to start the drying step. The moisture content of the curds was followed by weighing either the whole plate regularly or by determining the moisture content by drying and weighing of individual samples. When the curds reached a moisture content of 40-45% (typical for Gouda) the plates were sealed in a nitrogen atmosphere to prevent contamination with aerobic microorganisms. Ripening was subsequently performed at 17 ⁰ C.

Example 2. Production of Gouda-type cheese according to the protocol of Example 1, employing (1) different starter cultures without additional adjunct cultures, and (2) different adjunct cultures in the absence of a starter culture.

In one variant of the protocol described according to Example 1 , the adjunct cultures were absent and the concentration of Bos in the cheese milk composition was 1.5%. In another variant, Bos was replaced by a mixture of HBlO and Bos in a weight ratio of 30/70. The total concentration of the mixture of starter cultures in the cheese milk composition was 1%.

In yet another variant, the cheese milk composition contained the starter culture APS at a concentration of 1%.

In yet another variant, Bos was replaced by HBlO. The concentration of the starter culture HBlO in the cheese milk composition was 1%.

In another variant the cheese milk composition contained a culture of Lactobacillus acidophilus (T72) at a concentration of 1%.

Herein, all cultures except HBlO were obtained from CSK Food Enrichment; HBOlO is equivalent to MG5267(pNZ5519) as described in Bachmann et al. 2007, Appl Environ Microbiol. 2007 Jul;73(14):4704-6.

Example 3. Analysis of cheese obtained according to Examples 1 and 2.

After ripening at 17 ⁰ C for 7 and 14 days, the cheeses according to Examples 1 and 2 were analyzed for volatiles in the headspace using GC-MS equipment after 7 and 14 days, respectively. The results showed that typical key flavour compounds of cheese could be identified in all cheese varieties produced.

The results obtained for the cheeses according to Example 1 showed that the GC-MS profiles obtained were different allowing to distinguish between the two different adjunct cultures.

Analysis results for the cheeses obtained according to Example 2 are separately shown in Table 1 ; as a control, normally manufactured Gouda cheese was taken along.

Thus, for the cheeses obtained according to Example 2, after one week of ripening (at

17 ⁰ C), a change in the flavor profile could be observed; the observed changes are in agreement with literature.

Comparison to the control sample (of commercially produced young Gouda cheese) shows that the same flavor compounds are being identified but they are present in lower concentrations in the cheeses produced according to the present invention - this is expected because the latter were only 1-2 weeks of age.

A cluster analysis of multiple samples of the different cultures from above showed that the 4 different starter cultures used could be clearly distinguished from each other.

Table 1. Headspace GC-MS analysis of cheese obtained according to Example 2 (Gouda protocol), expressed as surface area (arbitrary units) for the specified key flavour components. The control sample was an equivalent amount of cheese (170mg) of a commercially produced young Gouda cheese.

APS (14 days) 76503 210328 518 7066 362 2677 111 2606 53981

Furthermore, the cheeses produced using HBlO were screened for luciferase activity for in situ analysis of gene expression.

Example 4. Protocol for preparing Cheddar-type cheese, using the method according to the present invention.

For the production of a Cheddar type cheese, pasteurized standardized cow's milk with a fat content of 3.6% and a protein content of 2.5% was used. The strains Lactococcus

lactis ssp. lactis and Lactococcus lactis ssp. cremoris were precultured in 10 ml of milk at 30 ⁰ C for 18 hours.

Kalase (150 IMCU/ml) (ex CSK Food Enrichment, Ede, The Netherlands), calcium chloride solution (33% wt/vol. in water) and the pre-incubated starter cultures were added, with brief mixing, to a volume of standardized cow's milk, for obtaining a cheese milk composition comprising a final concentration of 1.2 mM CaCl ₂ , 1% of a fully grown L. lactis ssp lactis culture, 0.5% of a fully grown L. lactis ssp. cremoris culture and 0.23 g Kalase per litre of cheese milk composition.

Each of the 96 wells of a microtitre plate (Greiner Bio-One master block 96 well, #78027x) were then filled with 1.7 ml of the mixture obtained above. The microtitre plate was then sealed with a cap mat to avoid evaporation of milk, spillage and contamination and the microtitre plate was incubated at 30 ⁰ C. After 55 min, the cheese milk composition had coagulated, and the curds thus formed were subsequently cut using a device comprising 96 pin-shaped bars, each of dimensions 1x50 mm. Cutting was performed by allowing each of the pin-shaped bars to penetrate the coagulum. Each of the pin-shaped bars was forced to produce a rotary movement in the horizontal plane at a speed of approx. 1 rotation per second for 10 mins. During the cutting step, the bottom side of the pin-shaped bars remained within a distance of between 100-3000 micrometers from the bottom of each cheese vat. After cutting the curds, the contents of the cheese vats were "tumbled" by sealing the plates with a capmat and placing the plate on a tumbler for -60 min. at about 20 rotations per minute. During this -60 minutes of tumbling the temperature was increased with ~1°C per 6 minutes until a temperature of 39°C is reached. Subsequently tumbling is continued at 39°C for another 60 minutes.

In the next step the curds were allowed to settle to the bottom for 10 minutes followed by centrifugation at 800 g for 5 minutes and removal of the whey. Then the plates were incubated again at 30 ⁰ C for 90 minutes. After the incubation period 9 μl of a 20% NaCl solution were added to each well (calculations see Gouda protocol) which should result in a salt concentration of 1.6% per dry matter cheese. However, it turned out that salt was lost during the next step, which is typical for the Cheddar protocol (i.e. the

completion of the separation of the curds from the whey). To compensate for this loss, it turned out that an additional amount of 5 μl of the 20% NaCl solution had to be added to each well in order to obtain the desired salt concentration in the cheese before ripening.

To completely separate the curds from the whey, the microtitre plate was centrifuged at 3000 rpm (1865 g) for 60 min at 30 ⁰ C. The supernatant was removed by decanting the microtitre plate and placing it upside down on a sterile tissue for 30 min. Then the plate was sealed using a gas and water permeable membrane which is impermeable to unwanted bacteria (Breathseal Greiner Bio-One #676050) and placed in a climate chamber set to 17 ⁰ C and 70% relative humidity in order to start the drying step. The moisture content of the curds was followed by weighing either the whole plate regularly or by determining the moisture content by drying and weighing of individual samples. When the curds reached a moisture content of 35-37% (typical for Cheddar) the plates could be sealed in a nitrogen atmosphere to prevent contamination with aerobic microorganisms. Ripening could subsequently be performed at 12 ⁰ C.

Example 5. Another production run of Gouda-type and Cheddar-type cheeses, using a standard 96 well microtitre plate.

In the present Example, the wells in a standard 96 deep well microtitre plate are used as an array of cheese vats that can be addressed individually with cultures. The stirring device having 3mm thick pin-shaped bars was employed. In each well an individual cheese is produced from as little as 1.7 ml of milk. In the present Example, such individual cheeses are being referred to as MicroCheeses.

Material and Methods

Milk, enzymes, starter cultures, culturing conditions

The milk used for manufacturing MicroCheese was standardized bovine milk with a fat content of 3.54%, protein content of 3.38% and a lactose content of 4.46%. It was pasteurized at 72.5°C for 9 seconds. To allow the manufacturing of cheese on different

days with the same batch of milk, aliquots were shock frozen by pouring milk directly into liquid nitrogen. Subsequently it was stored at -40 ⁰ C until usage. Starter cultures for Gouda type cheese, FRl 8 and APS, were obtained from CSK food enrichment (Ede,

The Netherlands). Starter cultures for Cheddar type cheese, Choozit™ RA21 LYO 250 DCU (named RA21 throughout this paper) and Choozit™ FLAV54 LYO 5D (named Flav54 throughout this paper), were obtained from Danisco (Copenhagen, Denmark). Because of the small quantities of starter culture needed the concentrated cultures were not added directly to the milk, but the cultures were pre -cultured in sterilized reconstituted skimmed milk powder (Promex Spray 1% skimmed milk powder; Friesland foods butter, Lochem, The Netherlands). For pre-culturing the starter cultures were grown at the following conditions: 20 hours at 20 ⁰ C for FRl 8, 16 hours at 37°C for APS, 18 hours at 30 ⁰ C for RA21 and 16 hours at 37°C for FLAV54. Determination of bacterial colony forming units in cheese was done by dissolving cheese in 2% sodium-citrate and subsequent plating on M 17 agar (Merck, Darmstadt, Germany) supplemented with 0.5% lactose. As a control sample for GC-MS and HPLC analysis a young Gouda type cheese (Jonge Beemster) was taken along.

Cheese manufacturing and ripening Gouda type cheese For cheese manufacturing the frozen milk was thawed and subsequently heated to 30.5 ⁰ C. The milk was supplemented with renneting enzyme (Kalase, CSK food enrichment, Ede, The Netherlands) at a concentration of 230 μl/liter milk. Furthermore 400 μl of a 33% w/v CaCl2 solution was added per liter milk. As a starter culture one percent of an FRl 8 pre-culture was added to the milk. For MicroCheeses produced with an adjunct culture 2.5% of an APS pre-culture were added to the milk. Following inoculation the wells of a 2 ml deepwell microplate (Greiner, Alphen a/d Rijn, The Netherlands) were filled with 1.7 ml of milk each and the plates were sealed with a capmat (Greiner, Alphen a/d Rijn, The Netherlands) and incubated at 30.5 ⁰ C. Separately 50 ml of the inoculated milk were incubated at the same temperature to monitor milk coagulation. After 45 minutes the cutting of the curds was started, using a custom made stirring device. This stirring device is made from stainless steel and consists a plate with a handle at the top and 96 pins attached to the bottom. The pins are 3 mm in diameter and 45 mm long, and they are aligned in a way so they are

positioned precisely in the middle of each well if the stirring device is placed in a 96- well microplate. The stirring device is used for cutting and stirring of the curds. Cutting of the curds was done by slow horizontal and vertical movements of the stirring device through the curds. This was followed by stirring of the curds and the two steps together took 20 minutes. During these 20 minutes cutting/stirring for 20 seconds and resting for 3 minutes were alternated. After the stirring, the curds where allowed to rest for 5 minutes and then the plates were sealed again and centrifuged at 466 g for 5 minutes to slightly compact the curds. Subsequently 680 μl of whey were removed from each well and replaced with 620 μl of sterile tab water. Throughout cutting, stirring and centrifuging the plates were kept at 30.5 ⁰ C. The addition of the washing water, which was heated to 45°C, brought the temperature in the wells to approximately 36°C. After the addition of the water the plates were placed in a water-bath tempered at 35.5°C and the curds were resuspended with the stirring device. This was followed by a 40 minutes incubation period at 35.5°C and regular stirring as described above. Then the plates were incubated at the same temperature for another 20 minutes without stirring. Whey was removed from the curds by centrifuging the microplates at 4800 g for one hour at 30 ⁰ C. The supernatant was discarded by decanting the plate and keeping the plate upside down on a tissue for 15 minutes. Finally the plates where covered with a Breathseal (Greiner; Alphen a/d Rijn, The Netherlands) and placed in a controlled climate stove at 30 ⁰ C and 30% relative humidity. After over night inoculation 17 ul of a 20% sodium chloride solution (w/v) was added to each well which, as calculated according to the Van Slyke and Price formula, should give approximately 3% salt in dry matter in the cheese. The evaluation of moisture content of the MicroCheeses was monitored by calculations based on the total weight of the cheese in each plate and by taking out individual samples and determining the moisture content as described below. After cheeses reached the target moisture content of 42-44% the plates were sealed in 0.8 atmospheres of 100% nitrogen. The cheeses were left for ripening at 17°C for 1 and 6 weeks before further analysis.

Cheddar type cheese

The ingredients for cheddar type cheese, milk, Kalase and CaCl2 were used in concentrations identical to Gouda type cheese. As starter culture for Cheddar we used 1% of a RA21 pre-culture. In case of the addition of an adjunct we added, next to

RA21, 0.75% of Flav54 pre-culture. After inoculation, the plates were sealed and incubated at 30 ⁰ C for 45 minutes. The coagulation status was assessed as described above. The coagulated milk was cut and stirred with the cutting device for 10 minutes as described above. Subsequently stirring was continued and the temperature was raised with 1°C per six minutes to 39°C. The increase in temperature in this protocol is relatively slow which ensures equal temperature increase throughout the plate. Then stirring was continued for another 60 minutes at 39°C as described above. After the stirring was finished the curds were allowed to rest for 10 minutes which was followed by centrifugation of the plates at 466 g for 5 minutes at 30 ⁰ C. After centrifugation the whey was removed and the sealed plate was incubated for another 90 minutes at 30 ⁰ C. In the following step 30 ul of a 30% (w/v) sodium chloride solution was added to each well and the plate was kept at 30 ⁰ C for another 5 minutes. Subsequently the plates were centrifuged at 4800 g and 30 ⁰ C for 60 minutes. The supernatant was removed by decanting the plate and placing it upside-down on a dry tissue for 15 minutes. Next the plates were sealed with a Breathseal (Greiner; Alphen a/d Rijn, The Netherlands) and to reduce moisture content in the cheese they were incubated in a controlled climate stove at 30 ⁰ C and 30% relative humidity for 16 hours. The target moisture content was 35-37% and it was measured and monitored as described above. The cheeses were ripened at 17°C for 1 and 6 weeks.

Moisture, Salt, Fat and pH determination

Moisture analysis was performed according to the dutch norm NEN 3755 (Nederlands centrum van normalisatie, Delft, The Netherlands) with the alteration that for each determination only approximately 170 mg of cheese (one MicroCheese) were used. The analysis of the salt content of MicroCheese was performed potentiometrically according to dutch norm NEN 3762 (Nederlands centrum van normalisatie, Delft, The Netherlands) with the alteration that for each determination only -170 mg of cheese were used. The fat content in cheese was determined butyrometically according to the dutch norm NEN 3758 (Nederlands centrum van normalisatie, Delft, The Netherlands). For each analysis of the fat content 15 MicroCheeses were pooled, to have sufficient material (>2 grams) for the butyrometric fat determination. The pH values were determined with a 3 mm pH electrode (BioTrode, Metrohm; Herisau, Switzerland) which was used according to the manufactures instructions. For pH measurements in

MicroCheese this electrode was pressed approximately 2 mm into the MicroCheese. All yield calculations were performed according to the Van Slyke and Price formula.

Proteolysis Determination of proteolysis in MicroCheese samples was essentially performed as described in Recio, L, and S. Visser ( 1999), "Two ion-exchange chromatographic methods for the isolation of antibacterial peptides from lactoferrin. In situ enzymatic hydrolysis on an ion-exchange membrane", J Chromatogr A 831:191-201. In short, analytical reversed phase chromatography (RP-HPLC) was carried out by using two M 6000A pumps in combination with a high sensitivity accessory block (Waters), an ISS- 100 injector (Perkin-Elmer, Uberlingen, Germany), a Waters Model 680 gradient controller and a Kratos 783 detector (Kratos Analytical, Ramsey, NJ, USA). A 250 x 4.6 mm Widepore Cl 8 column (Bio-Rad Laboratories, Richmond, CA, USA) was used with a Cl 8 cartridge (Bio-Rad) as a guard column. Column temperature was 30 ⁰ C throughout. The equipment was linked to a data acquisition and processing system (Turbochrom, Perkin-Elmer). Solvent A was a mixture of acetonitrile-water- trifluoracetic acid (TFA) (20:980:1, v/v/v) and solvent B contained the same components (900:100:0.8, v/v/v). Components were eluted with a linear gradient of solvent B in A. The absorbance of the eluent was monitored at 220 nm and 280 nm. Injection volumes were 100 μl.

GC-MS analysis

For volatile analysis an Ultra fast GC-MS setup was used. Each MicroCheese was transferred to a 1.5 ml headspace vial. The samples were automatically extracted and injected into the GC-MS by a Combi Pal Autosampler (CTC Analytics AG; Zwingen, Switzerland). Vials were heated to 60 ⁰ C before 500 μl of the headspace volatiles were injected into the GC column (UFM RTX 200 10m x 0.14mm; Thermo Fisher Scientific, Inc.; Waltham, USA). The initial temperature of the GC column was held for 0.4 minutes at 20 ⁰ C and subsequently it was raised at 200°C/min. to the final temperature of 250 ⁰ C on which it was held for 0.5 minutes. The total run time, including cooling, took approximately 5 min. Mass spectra were recorded by a Thermo plus Time of Flight mass spectrometer (Thermo Fisher Scientific, Inc.; Waltham, USA). The detection of mass spectra was performed with an ionization energy of 70 ev and a

scanning rate of 25 scans/s. The detected m/z ratio ranged from 35 - 350. Peak identification was done using the NIST MS search program version 2.0. Quantitation of Peak areas was performed using XCalibur 1.4SR1 (Thermo Fisher Scientific, Inc.; Waltham, USA ).

CSLM microscopy

Imaging was performed using a LEICA TCS SP Confocal Laser Scanning Microscope in the fluorescence, single photon mode. The set-up was configured with an inverted microscope (model LEICA DM IRBE) and an Ar/Kr laser. The objective lens used was a 63x/NA1.2/Water immersion/PL APO. Nile Blue A (N0766; Sigma, Zwijndrecht, The Netherlands) was used to stain the inclusion of lipids in the cheese matrix.

Statistical analysis

Clustering and analysis of correlation of the compound data of all GC-MS profiles with ripening periods and starter composition was performed using Random Forest, cf. Breiman, L (2001) "Random Forests", Machine Learning 45:5-32.

Results

For the validation of the model we prepared 4 different types of cheese Gouda with and without and adjunct culture and Cheddar with and without an adjunct culture. For each of the 4 different types of cheese we manufactured, 2 x 96 MicroCheeses on two different days, and multiple samples were monitored and analyzed for acidification rates, moisture content, salt concentration, proteolysis and volatile flavor compounds. The analysis of proteolysis and volatile flavor compounds was performed on Microcheese ripened for one and for six weeks. The amount of rennet, CaCl2 and starter culture added to the milk as well the temperature regime followed throughout cheese manufacturing were kept identical with conventional cheese making protocols. For the cutting and stirring of the curds the special device was used as described above, as developed for cutting the curds and allowing for the simultaneous stirring of the contents of each of the cheese vats provided by each of the wells of a 96 -wells microtitre plate. The cutting of the curds results in a curd size of 0.2 - 0.4 mm in diameter for Gouda type cheese and 0.9 - 1.1 mm in diameter for cheddar type cheese. The small curd size is inherent to the system, with the size of the cheese vats being

7x7x40 mm. The pressing of the curds applied in industrial scale cheese manufacture is replaced by centrifugation. The addition of salt to either Gouda or Cheddar type cheese is not done conventionally - i.e. by submerging the cheese in brine or kneading the salt into the curds respectively - but by adding a defined amount of brine to the cheese which is being absorbed. Eventually the plates were sealed in a nitrogen atmosphere to prevent fungal contamination and the cheeses are allowed to ripen at 17°C. To avoid positional effects on the microplate caused by temperature gradients, the initial filling of the wells was done with milk which was already at the incubation temperature of the coagulation step and subsequently it was carefully kept at temperature. To assess if there is a positional effect on the acidification profile we measured the pH in different positions covering all areas of a microplate 4, 5.5 and 24 hours after beginning of cheese manufacturing which we define as the point of inoculation of milk with enzymes and starter culture. We compared the pH values measured in borderwells, to the pH values measured on the inside of the plate. For none of the assessed time points we could find a significant difference indicating homogenous conditions during cheese manufacturing.

The acidification rates for Gouda as well as for Cheddar type cheese show acceptable standard deviations within one experiment and between experiments carried out on different days. The moisture content for both, Gouda and Cheddar were between 45% and 50% after the last centrifugation step of each protocol. To adjust the moisture content of the cheeses the plates were incubated in the controlled climate stove. The moisture content after 40 hours of incubation in the climate stove were on average 42.8% and 44.2% for the Gouda type cheeses produced on different days. These moisture contents were within our targeted values and cheeses were sealed in a nitrogen atmosphere to avoid fungal contaminations and incubated for ripening at 17°C until further analysis. The moisture content after 40 hours of incubation in the controlled climate stove was on average between 40.6% and 41.6% for Cheddar type cheeses produced on different days, which was not yet within target for this type of cheese. However, after further incubation of the mictrotitre plate(s) holding said Cheddar-type cheeses in a climate stove for 24 hours the average moisture content dropped to our target values of 35% - 38% and plates were then sealed in a nitrogen atmosphere and ripened at 17°C until further analysis.

As mentioned above, the addition of salt to the MicroCheese is done by adding a defined amount of sterile brine to each cheese which is being absorbed. The amounts of sodium chloride added were calculated to give the concentration as found in the cognate normal scale cheese protocol. For a Gouda type cheese this is easily predictable because the total amount of the added brine remains in each well and hence the Van Slyke and Price formula can be used straight away. For Cheddar the Van Slyke and Price formula was also used as a starting point for estimating the total amount of brine to be added, but the amount thus calculated was enhanced by a predetermined amount (which may be determined by routine trial experiments). This isbecause the salt is added before the last centrifugation step, after which excess whey is still being removed. During the removal of the excess whey also sodium chloride is being lost. This loss was compensated by increasing the initial amount of sodium chloride added. The analysis of our cheeses shows a NaCl concentration of approximately 3% (in dry matter) for Gouda and Cheddar type cheese. We also pooled Gouda type MicroCheeses and determined the fat content. The results showed that the fat percentage in dry matter ranged between 39.9 and 40.5%. Based on this results we calculated the fat loss in our system to be approximately 19% which is significantly higher than the 7% normally seen in industrially manufactured cheese. This fat loss is probably due to the small curd size and the centrifugation steps in our protocol (see below for discussion). Knowing the fat loss in our system and the casein content of the milk we calculated the expected cheese yield to be 10.82%. The actual yield obtained in our system for gouda type cheese was 10.45%. To investigate the microstructure of the MicroCheese we performed CLSM microscopy. Gouda type MicroCheese and conventionally manufactured cheese were stained with Nile Blue to visualize the lipid fraction enclosed in the cheese matrix. For both samples the images show similar results. Coalescence of the fat globules is clearly present in our model system, indicating similarities on a microstructure level between the two samples. CLSM microscopy was performed on three individually manufactured MicroCheeses with similar results. HPLC analysis of MicroCheeses show that proteolysis within and between different experiments is reproducible. Peaks were identified based on previously published chromatograms (Exterkate, F. A., C. Slangen, and R. J. Siezen (2001) "Effect of genetically modified Lactococcus lactis cell-envelope proteinases with altered specificity on the course of casein degradation under cheese conditions", International

Dairy Journal 11:363-371) and the results show for Gouda type cheese that α-Sl casein clearly decreased between week 1 and week 6 of cheese ripening. At the same time one of its breakdown products, α-Sl-I casein increased. For 6 weeks old MicroCheese the trend regarding α-Sl-I clearly went towards the expected breakdown product. The measured concentrations of para-casein, α-S2 casein and β-casein were very well comparable between 6 weeks old MicroCheese and our control cheese. The breakdown product of β-casein, β-casein-I was not detectable in one week old MicroCheese, but was found in comparable amounts to our control cheese after 6 weeks of ripening. Furthermore flavor volatiles were determined using an ultrafast GC-MS setup. The results showed that we were able to identify typical cheese flavour compounds such as acetic acid, acetoin, butanoic acid, diacetyl, ethylbutyrate, ethylhexanoate, ethyloctanoate, 3-methyl-butanal, hexanoic acid, limonene, 2-nonanon, pentanone, pentenal, 2-propanon and undecanone. The areas of all identified peaks were determined and the data was subjected to detailed statistical analysis. Unsupervised clustering showed that the replicate samples clustered together very well. Consequently, a full Random Forest classifier (classifying both ripening periods and starter compositions) based on the compound data showed a low out of box (OOB) error estimate of approximately 5%. A decision tree classifier built with the "tree" package in R and based on the four compounds with the strongest predictive power, being acetic acid, acetoin, hexanoic acid and butanoic acid had only a slightly higher error rate of 6%. It visualizes the relations between cheese classes and compound levels, and shows putative interactions between compound levels. The data clearly allows distinguishing between the four different types of cheeses manufactured. Also the difference between samples with either 1 or 6 weeks ripening time could be clearly distinguished. In general the samples ripened for 6 weeks showed higher concentrations of most flavor compounds which is consistent with what is known by the person skilled in the art for full-size industrially produced cheese. No difference could be shown between the same types of cheeses produced on different days, confirming good reproducibility of our results. Technical replicates of the GC-MS analysis (the same cheese analyzed twice) gave highly reproducible chromatograms.

Discussion

The present results demonstrate again a protocol which allows the manufacturing of individual cheeses from as little as 1.7 ml of milk, or even less. It is effectively a downscaling of the cheese manufacturing process to the microplate format with the current setup one person can handle almost 600 cheeses per day. The results show that with the developed protocol key parameters in cheese, such as moisture and salt content could be adjusted to particular target values. The bacterial acidification rate in the cheese was found to be similar to normal cheese manufacturing. Furthermore it was shown that the variations of key parameters are relatively well reproducible within and between production runs. We tested for border- well effects on the microplate but monitoring of acidifications profiles in microcheeses in all positions on the plates showed a homogenous distribution. The systems can be operated with high reproducibility. Although the variation in moisture content and the end pH between the different production runs in the present Example might reside somewhat above the variation seen during conventional cheese manufacturing, the vast number of results which could be generated allowed the overall standard deviation to be reduced to acceptable values. It is further noted that the variation in moisture content may largely be explained by the measuring error, which is increased compared to the standard procedure due to the small amount of cheese available per sample.

The addition of the right amount of salt to a MicroCheese could be conveniently performed by adding a defined amount of brine which is being absorbed by the cheese. This allowed a very precise control of salt addition especially for Gouda type cheese. The largest difference of microcheese compared to normal cheese making is the reduced fat content, which is most likely a consequence of the smaller curd size and/or the centrifugation steps in our protocol. Now that the extra loss of fat is known for the current set-up in the present Example, this problem may be easily solved by the addition of extra (milk) fat to the cheese milk at the beginning of the cheese making process. The microstructure of the cheeses was assessed using CSLM microscopy and the results clearly show coalescence of fat particles, which are very well comparable to industrially produced cheese. To investigate if bacterial activities leading to textural changes and the formation of flavor molecules are similar to conventionally manufactured cheese we assessed proteolysis and flavor profiles of Microcheeses.

HPLC analysis showed that the breakdown of milk proteins is in many aspects similar to our control sample. Volatile flavor compounds in Microcheeses were determined after 1 and 6 weeks of cheese ripening and the outcome clearly allows distinguishing between the different samples. The presented data shows that MicroCheeses manufactured in the microtitre plates resemble most key properties of conventionally manufactured cheeses.